Among the various photovoltaic technologies, perovskite solar cells have attracted extensive attention because of their low production cost and rapidly improving power conversion efficiency. Since the first report of 3.8% efficiency, single-junction perovskite solar cells have advanced to 27.0%, becoming comparable to conventional silicon-based cells. To further advance perovskite solar cells, the selection of suitable chemical substances as passivation materials is a key strategy for improving performance. Polymer materials have been widely studied in perovskite solar cells because of their diverse and tunable physical and chemical properties. In particular, polyurethane (PU) can endow perovskite films with self-healing ability, optimize surface morphology, and passivate grain boundaries. However, the use of PU as a buried-interface passivation material for perovskite solar cells has rarely been explored. In this work, I synthesized PU containing boronic ester structures and used it as a buried-interface passivation material to construct inverted perovskite solar cells. The introduction of PU reduced the surface potential of the hole transport layer, improved energy-level matching, and created multiple chemical interactions with the perovskite, thereby passivating the buried bottom surface. The PU-modified device reached a power conversion efficiency of 22.68%.
In the following discussion, I present the experimental design, photovoltaic performance, surface potential characterization, morphology and crystallization behavior, chemical passivation mechanisms, and the overall conclusions for inverted perovskite solar cells modified with PU. I also summarize the key parameters using tables and equations so that the relationships among composition, interface chemistry, and device performance can be clearly compared.
The buried interface between the hole transport layer and the perovskite layer in an inverted device not only influences charge transfer and extraction but also determines the film quality of the perovskite. Therefore, I focused on modifying this buried interface with a functional polymer. The polymer was designed to contain boronic ester groups as well as urethane and amine functionalities, because these groups can participate in hydrogen bonding, coordination, and Lewis acid-base interactions with perovskite components. Such interactions are expected to reduce interfacial defects and improve crystallization. To test this hypothesis, I first synthesized the polymer, then fabricated inverted perovskite solar cells with and without a PU interlayer, and finally characterized the optical, electronic, morphological, crystallographic, and chemical properties of the resulting films and devices.
The experimental work began with the preparation of the polymer. I used isophorone diisocyanate (IPDI), polycaprolactone diol (PCL-1000), a boronic ester-containing diol (HDB), and N,N-dimethyl-N,N′-bis(2-hydroxypropyl)-1,3-propanediamine (DPA) as the main reactants. The synthetic route involved a prepolymerization step, a chain extension step with HDB, and a final chain extension step with DPA. The reaction sequence can be summarized as follows:
$$
\text{IPDI} + \text{PCL-1000} \xrightarrow[\text{DBTDL}]{\text{80 }^\circ\text{C, 3 h}} \text{NCO-terminated prepolymer}
$$
$$
\text{NCO-terminated prepolymer} + \text{HDB} \xrightarrow{\text{80 }^\circ\text{C, 2 h}} \text{boronic ester-containing prepolymer}
$$
$$
\text{boronic ester-containing prepolymer} + \text{DPA} \xrightarrow{\text{80 }^\circ\text{C, 2 h}} \text{PU}
$$
In a typical synthesis, I combined 3 g (13.5 mmol) of IPDI and 4.5 g (4.5 mmol) of PCL-1000 under mechanical stirring at 80 °C. After adding a dibutyltin dilaurate catalyst, the mixture was allowed to react for 3 h. Then, 0.91 g (2.5 mmol) of HDB was added, and the reaction continued for 2 h. Finally, 1.09 g (5.0 mmol) of DPA was added, and the reaction proceeded for another 2 h. During the reaction, anhydrous N,N-dimethylacetamide was added as needed to control viscosity. The final product was a PU containing boronic ester structures. The chemical structure of the PU contains urethane linkages, boronic ester groups, and tertiary amine moieties, which provide multiple sites for interaction with perovskite precursors and defect sites.
For device fabrication, I used indium tin oxide (ITO) substrates. The ITO substrates were sequentially cleaned with pure water and ethanol under ultrasonication for 20 min each and then dried with a nitrogen gun. The cleaned ITO was treated with ultraviolet-ozone for 20 min. A 0.5 mg/mL solution of MeO-2PACz in ethanol was spin-coated onto the ITO at 3000 rpm, followed by annealing at 100 °C for 10 min. Then, a PU solution in DMF was spin-coated onto the ITO/MeO-2PACz substrate at 3000 rpm, followed by annealing at 100 °C for 10 min. The perovskite precursor solution was prepared by dissolving 0.0110 g PbBr2, 0.0033 g MABr, 0.0195 g CsI, 0.0152 g MACl, 0.2399 g FAI, and 0.7468 g PbI2 in 1 mL of a DMF:DMSO = 4:1 mixed solvent. The precursor solution was deposited onto the PU layer by a two-step spin-coating method: first at 1000 rpm for 5 s, then at 5000 rpm for 40 s. Chlorobenzene was dropped 8 s before the end of the second step. The film was annealed at 110 °C for 20 min. A 3 mg/mL PEAI solution in isopropanol was spin-coated onto the perovskite layer at 4000 rpm. Subsequently, a 20 mg/mL PC61BM solution in chlorobenzene was spin-coated at 2000 rpm. Then, a 0.5 mg/mL BCP solution in isopropanol was spin-coated at 5000 rpm. Finally, a silver electrode of about 70 nm was thermally evaporated under vacuum.
The photovoltaic performance was measured using a solar simulator and a digital source meter. The current density-voltage (J-V) curves were recorded under simulated AM 1.5G illumination. The contact potential difference (CPD) of the films was measured by Kelvin probe force microscopy (KPFM). Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS) were used to analyze functional groups and elemental compositions. Scanning electron microscopy (SEM) was used to examine surface morphology. X-ray diffraction (XRD) was used to study crystal structure. Steady-state photoluminescence (PL) was used to evaluate radiative and nonradiative recombination. The power conversion efficiency (PCE) was calculated according to:
$$
\text{PCE} = \frac{V_{\text{OC}} J_{\text{SC}} \text{FF}}{P_{\text{in}}}
$$
where \(V_{\text{OC}}\) is the open-circuit voltage, \(J_{\text{SC}}\) is the short-circuit current density, \(\text{FF}\) is the fill factor, and \(P_{\text{in}}\) is the incident power density, typically 100 mW/cm² under standard test conditions.
I first optimized the concentration of PU used as the buried-interface layer. Different concentrations of PU solution were deposited on the hole transport layer, and complete devices were fabricated. The J-V curves showed that the best performance was obtained with 0.1 mg/mL PU. The photovoltaic parameters of the best devices with different PU concentrations are summarized in Table 1. The control device without PU exhibited a PCE of 21.33%, with \(V_{\text{OC}}\) = 1.104 V, \(J_{\text{SC}}\) = 23.04 mA/cm², and FF = 83.90%. When 0.05 mg/mL PU was used, the PCE increased to 22.19%, with \(V_{\text{OC}}\) = 1.119 V, \(J_{\text{SC}}\) = 23.31 mA/cm², and FF = 85.09%. The best performance was achieved at 0.1 mg/mL PU, giving a PCE of 22.68%, \(V_{\text{OC}}\) = 1.132 V, \(J_{\text{SC}}\) = 23.38 mA/cm², and FF = 85.71%. At a higher concentration of 0.2 mg/mL, the PCE decreased slightly to 22.30%, with \(V_{\text{OC}}\) = 1.135 V, \(J_{\text{SC}}\) = 23.18 mA/cm², and FF = 84.78%. These results indicate that an appropriate amount of PU improves charge extraction and reduces interfacial recombination, while excess PU may increase series resistance or disturb the interface.
| PU concentration (mg/mL) | \(V_{\text{OC}}\) (V) | \(J_{\text{SC}}\) (mA/cm²) | FF (%) | PCE (%), best |
|---|---|---|---|---|
| Control | 1.104 | 23.04 | 83.90 | 21.33 |
| 0.05 | 1.119 | 23.31 | 85.09 | 22.19 |
| 0.1 | 1.132 | 23.38 | 85.71 | 22.68 |
| 0.2 | 1.135 | 23.18 | 84.78 | 22.30 |
The dark J-V curves provided further evidence for defect reduction. In the dark, the current is mainly carried by leakage pathways associated with defects. The PU-modified device exhibited a lower dark current density than the control device. Since defects create additional conductive channels and give rise to dark current, the reduced dark current density indicates a lower density of defect states. Therefore, PU modification effectively reduces defects in the device. To quantify this effect, I also fabricated single-hole devices and measured the trap-filled limit voltage (\(V_{\text{TFL}}\)). The control device had a \(V_{\text{TFL}}\) of 1.17 V, while the PU-modified device had a \(V_{\text{TFL}}\) of 1.00 V. The trap density can be estimated from:
$$
V_{\text{TFL}} = \frac{q N_t L^2}{2 \varepsilon \varepsilon_0}
$$
where \(q\) is the elementary charge, \(N_t\) is the trap density, \(L\) is the film thickness, \(\varepsilon\) is the relative dielectric constant, and \(\varepsilon_0\) is the vacuum permittivity. Because \(V_{\text{TFL}}\) is proportional to \(N_t\), the decrease from 1.17 V to 1.00 V corresponds to a reduction in trap density by a factor of approximately:
$$
\frac{N_{t,\text{PU}}}{N_{t,\text{control}}} = \left(\frac{1.00}{1.17}\right)^2 \approx 0.73
$$
Thus, the PU-modified device has about 27% fewer traps than the control device. This reduction is consistent with the lower dark current and improved photovoltaic parameters. To evaluate stability, I stored unencapsulated devices in a nitrogen glovebox at 25 °C for 43 days. The control device retained 72% of its initial efficiency, while the PU-modified device retained 94% of its initial efficiency. This improved stability suggests that the PU interlayer passivates defects and suppresses degradation pathways at the buried interface.
| Device | \(V_{\text{TFL}}\) (V) | Relative trap density \(N_t\) | Retained PCE after 43 days (%) |
|---|---|---|---|
| Control | 1.17 | 1.00 (reference) | 72 |
| PU-modified | 1.00 | 0.73 | 94 |
Surface potential characterization was performed by KPFM to understand how PU modifies the hole transport layer. I compared the ITO/MeO-2PACz substrate with the ITO/MeO-2PACz/PU substrate. The average contact potential difference decreased from 72.7 mV to 54.8 mV after PU introduction. The relationship between the contact potential difference and the work function can be expressed as:
$$
V_{\text{CPD}} = \frac{\Phi_{\text{tip}} – \Phi_{\text{sample}}}{e}
$$
where \(\Phi_{\text{tip}}\) is the work function of the tip, \(\Phi_{\text{sample}}\) is the work function of the sample, and \(e\) is the elementary charge. A decrease in \(V_{\text{CPD}}\) indicates an increase in the sample work function, which corresponds to a downward shift of the Fermi level. This shift is beneficial for energy-level matching with the perovskite layer and can enhance hole transport. Therefore, the PU interlayer not only passivates defects but also tunes the interfacial electronic structure of the hole transport layer.
The morphology of the perovskite films was examined by SEM. The control film, deposited on ITO/SAM, showed relatively small grains. After PU modification, the perovskite film exhibited noticeably larger grains. The increased grain size can be attributed to strong interactions between PU and \(\text{Pb}^{2+}\) ions, which slow down the crystallization rate of the perovskite film. A slower crystallization rate allows nuclei to grow into larger grains and reduces the number of grain boundaries. Since grain boundaries are often rich in defects and act as nonradiative recombination centers, larger grains are generally desirable for efficient perovskite solar cells.
To investigate the effect of PU on crystallization kinetics, I performed XRD measurements. Compared with the control perovskite film, the PU-modified film showed a significantly stronger characteristic peak for the (001) plane. At the same time, the intensity of the PbI2 peak decreased significantly. The intensity of the PbI2 peak is related to the degree of phase conversion. A stronger PbI2 peak usually indicates incomplete conversion, while a weaker PbI2 peak indicates more complete conversion to the perovskite phase. Therefore, the reduced PbI2 peak intensity suggests that PU promotes the conversion of precursors into the perovskite phase. In addition, the PL spectra showed that the PU-modified perovskite film had a higher PL intensity than the control film, which implies reduced nonradiative recombination. The relationship between PL intensity and recombination can be described by:
$$
\text{PLQY} = \frac{k_{\text{rad}}}{k_{\text{rad}} + k_{\text{nonrad}}}
$$
where \(k_{\text{rad}}\) is the radiative recombination rate and \(k_{\text{nonrad}}\) is the nonradiative recombination rate. A higher PL intensity indicates that nonradiative recombination is suppressed, which is consistent with the reduced trap density and improved device performance. The overall morphological and crystallographic results are summarized in Table 2.
| Property | Control film | PU-modified film | Interpretation |
|---|---|---|---|
| SEM grain size | Smaller | Larger | Slower crystallization and reduced grain boundaries |
| XRD (001) peak | Reference | Stronger | Improved perovskite crystallinity |
| XRD PbI2 peak | Stronger | Weaker | More complete phase conversion |
| PL intensity | Lower | Higher | Reduced nonradiative recombination |
To understand the chemical passivation mechanism, I used XPS and FTIR to analyze the buried interface. The C 1s XPS spectrum of the control perovskite film showed three peaks, which can be assigned to C-C bonds, the \(\text{N}^+=\text{CH-N}\) group of \(\text{FA}^+\), and the \(\text{CH}_3\text{-N}\) group of \(\text{MA}^+\). After PU modification, a new peak appeared, corresponding to the C=O bond of PU. In addition, the binding energies of the \(\text{N}^+=\text{CH-N}\) and \(\text{CH}_3\text{-N}\) peaks shifted. These shifts indicate that \(\text{FA}^+\) and \(\text{MA}^+\) may form hydrogen bonds with the C-O groups of the urethane moieties in PU. The Pb 4f spectrum also changed. After PU modification, the two characteristic peaks of Pb 4f shifted to lower binding energy. This shift can be attributed to electron transfer from the lone pair electrons of the C=O groups in PU to \(\text{Pb}^{2+}\). Furthermore, the I 3d spectrum showed shifts after PU modification, which can be explained by hydrogen bonding between the -NH groups in PU and \(\text{I}^-\) ions. These multiple interactions are beneficial for passivating the bottom surface of the perovskite.
The XPS peak assignments and shifts are summarized in Table 3. The presence of C=O in the C 1s spectrum confirms that PU remains at the buried interface after device fabrication. The shifts in the FA+, MA+, Pb2+, and I- signals provide direct evidence for chemical bonding between PU and the perovskite components.
| XPS spectrum | Species | Assignment | Observed change after PU |
|---|---|---|---|
| C 1s | C-C | Reference carbon | No significant change |
| C 1s | \(\text{N}^+=\text{CH-N}\) | FA+ | Binding energy shift |
| C 1s | \(\text{CH}_3\text{-N}\) | MA+ | Binding energy shift |
| C 1s | C=O | PU urethane group | New peak |
| Pb 4f | \(\text{Pb}^{2+}\) | Pb 4f7/2 and Pb 4f5/2 | Shift to lower binding energy |
| I 3d | \(\text{I}^-\) | I 3d3/2 and I 3d5/2 | Binding energy shift |
FTIR measurements provided additional evidence for the interaction between PU and PbI2. I compared neat PU, a PU/PbI2 mixture, and a film obtained by annealing a PU solution in DMF. In the neat PU spectrum, the B-O vibration peak was present. In the PU/PbI2 mixture, the B-O vibration peak disappeared. This disappearance can be attributed to an interaction between the boronic ester groups and PbI2. To rule out the possibility that the B-O peak disappeared because of DMF dissolution or annealing, I also measured a PU film that had been dissolved in DMF and annealed. In that film, the B-O vibration peak remained. Therefore, the disappearance of the B-O peak in the PU/PbI2 mixture is specifically due to interaction with PbI2, rather than solvent or thermal effects. The FTIR results are summarized in Table 4.
| Sample | B-O vibration peak | Interpretation |
|---|---|---|
| Neat PU | Present | Reference |
| PU + PbI2 mixture | Absent | Interaction between boronic ester and PbI2 |
| PU in DMF, annealed film | Present | Solvent and annealing do not remove B-O |
On the basis of the XPS and FTIR results, I propose that the PU interlayer passivates the buried interface through several complementary interactions. These include hydrogen bonding between PU and organic cations, coordination between C=O groups and \(\text{Pb}^{2+}\), hydrogen bonding between -NH groups and \(\text{I}^-\), and interaction between boronic ester groups and PbI2. These interactions can be represented as:
$$
\text{PU} \cdots \text{FA}^+, \quad \text{PU} \cdots \text{MA}^+, \quad \text{PU} \cdots \text{Pb}^{2+}, \quad \text{PU} \cdots \text{I}^-
$$
$$
\text{R-B(OR’)}_2 + \text{PbI}_2 \rightarrow \text{R-B(OR’)}(\text{I}) + \text{Pb(OR’)I}
$$
Such multiple interactions are expected to reduce undercoordinated Pb2+ and halide vacancies, suppress nonradiative recombination, and improve the quality of the buried interface. The chemical passivation also explains the reduced trap density, lower dark current, enhanced PL intensity, and improved device stability. In addition, the boronic ester groups may introduce dynamic covalent bonding, which can contribute to the self-healing character of PU. Although the primary role of PU in this work is interfacial passivation, the presence of dynamic boronic ester bonds may help the film tolerate strain and recover from local damage.
The energy-level alignment at the hole transport layer/perovskite interface is another important factor. The KPFM results showed that PU reduces the surface potential of the hole transport layer, which increases the work function and shifts the Fermi level downward. A downward shift of the Fermi level can improve the energetic match between the hole transport layer and the perovskite, reducing the hole extraction barrier. The hole extraction process can be described by:
$$
J_{\text{hole}} = q \mu_p p E
$$
where \(J_{\text{hole}}\) is the hole current density, \(\mu_p\) is the hole mobility, \(p\) is the hole concentration, and \(E\) is the electric field. Improved energy-level alignment increases the driving force for hole extraction and reduces interfacial recombination. Therefore, the PU interlayer contributes to both chemical passivation and electronic optimization of the buried interface.
The combined effects of PU modification on inverted perovskite solar cells are summarized in Table 5. The table compares the control and PU-modified devices in terms of photovoltaic parameters, defect density, surface potential, morphology, crystallinity, recombination, and stability.
| Parameter | Control | PU-modified | Improvement mechanism |
|---|---|---|---|
| PCE (%) | 21.33 | 22.68 | Reduced defects and improved charge extraction |
| \(V_{\text{OC}}\) (V) | 1.104 | 1.132 | Suppressed nonradiative recombination |
| \(J_{\text{SC}}\) (mA/cm²) | 23.04 | 23.38 | Better crystallization and charge transport |
| FF (%) | 83.90 | 85.71 | Lower series resistance and fewer traps |
| \(V_{\text{TFL}}\) (V) | 1.17 | 1.00 | Lower trap density |
| Dark current | Higher | Lower | Fewer leakage pathways |
| Surface potential (mV) | 72.7 | 54.8 | Improved energy-level matching |
| PL intensity | Lower | Higher | Reduced nonradiative recombination |
| Stability after 43 days (%) | 72 | 94 | Passivated buried interface and suppressed degradation |
I also considered the possible influence of PU concentration on film formation. At low concentration, PU can uniformly modify the hole transport layer and provide sufficient interaction sites for passivation. At the optimal concentration of 0.1 mg/mL, the device achieves the best balance between passivation and charge transport. At higher concentration, excess PU may increase the insulating character of the interlayer or create an unfavorable morphology, leading to slightly lower \(J_{\text{SC}}\) and FF. This concentration-dependent behavior is consistent with the observed trend in Table 1. The optimal concentration is therefore a compromise between chemical passivation and electrical transport.
The role of the boronic ester structure deserves further discussion. Boronic ester groups are known to participate in dynamic covalent chemistry and can interact with metal ions and halides. In this work, FTIR showed that the B-O vibration peak disappears when PU is mixed with PbI2, indicating that the boronic ester groups interact with PbI2. This interaction may help anchor PU to the perovskite surface and may also passivate undercoordinated Pb2+ sites. Because the buried interface is difficult to access after device fabrication, such anchoring is important for long-term stability. The retention of 94% of the initial efficiency after 43 days in a nitrogen glovebox supports the idea that the PU interlayer stabilizes the buried interface.
The SEM and XRD results together suggest that PU also influences perovskite crystallization. The larger grains observed by SEM indicate that PU reduces the nucleation density or slows the crystal growth rate. The stronger (001) XRD peak indicates improved crystallographic orientation and crystallinity. The weaker PbI2 peak indicates more complete conversion of PbI2 to the perovskite phase. These effects are beneficial because they reduce grain boundaries and residual PbI2, both of which can act as defect sources. The PL enhancement further confirms that nonradiative recombination is reduced. Therefore, PU modification improves the perovskite film quality through both interfacial passivation and crystallization control.
The XPS results provide a molecular-level explanation for the passivation. The appearance of the C=O peak in the C 1s spectrum confirms the presence of PU at the buried interface. The shifts of the FA+ and MA+ signals indicate hydrogen bonding between the organic cations and PU. The shift of the Pb 4f peaks to lower binding energy indicates electron donation from PU to Pb2+, which can fill undercoordinated Pb2+ sites. The shift of the I 3d peaks indicates hydrogen bonding between -NH groups and I- ions, which can passivate halide vacancies. These multiple interactions are cooperative and lead to a more stable and less defective interface. The proposed chemical interactions are summarized in Table 6.
| PU functional group | Perovskite species | Interaction type | Expected effect |
|---|---|---|---|
| C=O (urethane) | \(\text{Pb}^{2+}\) | Coordination / Lewis base | Passivates undercoordinated Pb2+ |
| C-O (urethane) | \(\text{FA}^+\), \(\text{MA}^+\) | Hydrogen bonding | Stabilizes organic cations |
| -NH (urethane/amine) | \(\text{I}^-\) | Hydrogen bonding | Passivates halide vacancies |
| B-O (boronic ester) | \(\text{PbI}_2\) | Lewis acid-base interaction | Anchors polymer and passivates Pb2+ |
Overall, the experimental results demonstrate that PU containing boronic ester structures is an effective buried-interface passivation material for inverted perovskite solar cells. The PU interlayer reduces the surface potential of the hole transport layer, improves energy-level matching, promotes larger perovskite grains, enhances crystallinity, reduces PbI2 residues, and passivates defects through multiple chemical interactions. The best device achieved a PCE of 22.68%, with \(V_{\text{OC}}\) = 1.132 V, \(J_{\text{SC}}\) = 23.38 mA/cm², and FF = 85.71%. The PU-modified device also showed improved stability, retaining 94% of its initial efficiency after 43 days in a nitrogen glovebox. These findings suggest that polymer interlayers with boronic ester groups can simultaneously address interfacial defects, crystallization, and energy-level alignment in perovskite solar cells.
In conclusion, I synthesized a PU containing boronic ester structures and used it as a buried-interface passivation material to construct inverted perovskite solar cells. The introduction of PU decreased the surface potential of the hole transport layer, which is favorable for energy-level matching and hole transport. FTIR and XPS analyses showed that PU interacts with the perovskite through hydrogen bonding, coordination, and Lewis acid-base interactions, thereby passivating the buried bottom surface. SEM, XRD, and PL measurements showed that PU modification improves grain size, crystallinity, phase conversion, and radiative recombination. The optimized device achieved a power conversion efficiency of 22.68%, and the unencapsulated device retained 94% of its initial efficiency after 43 days in a nitrogen glovebox. I therefore conclude that PU with boronic ester structures is a promising passivation material for high-performance and stable perovskite solar cells.
For future work, several directions can be explored. First, the molecular weight and boronic ester content of PU can be systematically varied to optimize the balance between passivation and charge transport. Second, advanced interface-sensitive techniques can be used to further resolve the depth distribution of PU and its chemical states. Third, the self-healing behavior of PU can be tested under mechanical bending or thermal cycling, especially for flexible perovskite solar cells. Fourth, the compatibility of PU with other hole transport layers and perovskite compositions can be examined. Finally, the scalability of the PU interlayer deposition process should be evaluated for large-area perovskite solar cells. These studies may further accelerate the development of polymer-passivated perovskite solar cells with high efficiency and long-term stability.